turbine design and analysis for the j-2xengine turbopumps

.'

Turbine design and analysis for the J-2X Engineturbopumps

Bogdan Marcu I and Ken Tran2

Pratt & Whitney Rocketdyne, Canoga Park, CA, 91309

Daniel 1. Dorney3 and Preston Schmauch4

NASA Marshall Space Flight Center, Huntsville, AL

Pratt and Whitney Rocketdyne and NASA Marshall Space Flight Center are developing the advancedupper stage J-2X engine based on the legacy design of the J-2/J-2S family of engines which powered theApollo missions. The cryogenic propellant turbopumps have been denoted as Mark72-F and Mark72-0 forthe fuel and oxidizer side, respectively. Special attention is focused on preserving the essential flight-provendesign features while adapting the design to the new turbopump configuration. Advanced 3-D CFD analysishas been employed to verify turbine aero performance at current flow regime boundary conditions and tomitigate risks associated with stresses. A limited amount of redesign and overall configuration modificationsallow for a robust design with performance level matching or exceeding requirement.

I. Introduction

T he J-2X engine is planned to power the Ares-I crew launch vehicle's upper stage and the Earth departurestage of the Ares-V cargo launch vehicle. Building on the flight-proven heritage of the Apollo J-2

engines, the J-2X is a gas generator cycle engine employing a fuel and an oxidizer turbopump whose turbines areconfigured in series, with the fuel turbine upstream of the oxidizer turbine. The utilization of the flight-proven J-2and heritage J-2S turbopump designs allows for a significant reduction in the development schedule, costs and risks.Originally designed in the 1960's, the turbopumps are being refurbished to operate at a higher thrust level of294,000 Ibf by employing new, state of the art modeling and analysis tools to insure an exceptional level of safetyand flawless operation during the planned space missions. Figure 1 and Figure 2 show comparisons of the twoturbopump designs side by side, the heritage Apollo era design and current design.

The present report addresses the necessary design changes and the analysis carried for the J-2X turbines whichhave been denoted as Mark72-F and Mark72-0 for the fuel and oxidizer sides, respectively. Pratt and WhitneyRocketdyne's (PWR) turbopump design team has applied the first-time-through design process demonstrated on theMBXX Demonstrator Engine Fuel Turbopump [1, 2] operating jointly with the NASA Marshall Flight Space Center(MSFC) team in an efficient and synergistic manner. This paper will present and discuss 3-D unsteady CFD resultsobtained by the MSFC team employing the Phantom CFD analysis code which was run at the NASA AdvancedSupercomputing Division center at Moffet Field, California. Aerodynamic loading obtained from the CFD analysishave been analyzed for turbine performance by the Pratt and Whitney Rocketdyne (PWR) aerodynamic analysisteam, while steady and unsteady pressure data from the CFD solution have been applied on the ANSYS structuralmodels for stress and High Cycle Fatigue (HCF) analysis at PWR.

II. Numerical Procedure for J-2X turbines aerodynamic analysisThe code used to perform the simulations is PHANTOM, a code developed at NASA Marshall Space Flight

Center. The governing equations in the PHANTOM code are the three-dimensional, unsteady, Navier-Stokesequations. The equations have been written in the Generalized Equation Set format, enabling it to be used for both

I Principal Engineer, Ph.D., Senior AIAA Member2 Principal Engineer3 Team Leader, Associate Fellow AIAA4 Aerospace Engineer

IAmerican Institute of Aeronautics and Astronautics

092407

liquids and gases at operating conditions ranging from incompressible to supersonic flow [3]. The code employs asystem of overset O-grids and H-grids to discretize the flow field. The grids move to simulate blade motion. Amodified Baldwin-Lomax turbulence model is used for turbulence closure [4]. The code contains two options for thefluid properties. The first option is based on the equations of state, thermodynamic departure functions andcorresponding state principles constructed by Oefelein [5]. The second option is based on splines generated from theNIST Tables [6]. Message Passing Interface (MPI) protocols are used for parallel simulations. A detailed descriptionof the code/algorithm development, as well as its application to several turbine and pump test cases, is presented inRefs. 3 and 7.

Figure 1. BaselineJ-2 and currentJ-2X fuel turbopurnp

,. '.

Figure 2. Baseline]-2 and current J-2X oxidizer turbopurnp

ill. The J-2X Fuel Turbine Mark72-F

A. General configuration

The heritage J-2S fuel turbine is a two stage impulse supersonic turbine with most of the pressure drop occurringin the 1st vane producing supersonic Mach numbers at the 151 vane exit. The flow is accelerated in the nozzle vane tohigh supersonic Mach levels through a convergent-divergent channel resulting in supersonic regime even in therotating reference frame of the 151 blade. The nozzle discharges at 19.so angle and 1.18 Mach on the 1st bladerelative frame as set by the geometry of the 15t blade leading edge through the "Unique Incidence" theory [8]relating incidence to inlet Mach number. A summary of the turbine operating parameters at 294K thrust level isdocumented in Table I.

2American Institute of Aeronautics and Astronautics

092407

Pressure ratioRPMu/CoMean Diameter (in)

6.00300000.18610.5

Table 1: 294K Ibf thrust operating point of the fuel turbine.

The new J-2X fuel turbine flow path is illustrated in Figure 3. The turbine is almost identical to the heritagedesign except for the I st nozzle vanes. The latter are modified to enhance structural margin of the 1st rotor blades.The 1st vane number is changed from 34 to 37 to prevent a near resonant interference with the blade torsional modein the Campbell diagram (Figure 4). In addition, the heritage radially stacked vanes are leaned to minimize thedynamic load on the vane trailing edge. In fact, structural forced response analyses indicate that the interactionbetween the supersonic flow at the exit of the vane and the 1st rotor blade is very strong producing quite high levelsof alternating stresses on Ist blade and on the 1st vane trailing edge. The 1st vane is leaned by 20° against thedirection of rotation based on CFD optimization.

37 vanes 69 blades 57 vanes 59 blades 52 vanes

Figure 3. The J-2X Mark72-F turbine flow path


092407

4,---------------------------.,,,..----,

..

:.

Potential Interference :3

1;;

cJwa:u. 2awa:u.

2nd Bending

oo 0.2 0.4 0.6 0.8 1.2 1.4

RPMIRPMdeSign

Figure 4. J-2X Mark72-F fuel turbine 1st rotor Campbell Diagram

B. J-2X Fuel turbine analyses

The 3D Phantom code coupled with ANSYS fmite element forced response analyses are employed to insure thereliability of turbine components. At the begirming of the program the entire turbine was modeled using cyclicalsector models to conserve rurming and post processing time. The row-by-row blade count of 37-69-57-59-52 inFigure 3 was modeled as 35-70-56-63-49 which, taking advantage of cyclic symmetry, can be reduced to a piesector of 5-1 0-8-9-7. The blade geometries were then scaled to fit the cyclical sector model. Although this procedurewas found to be adequate for aerodynamic performance predictions, it was found through further investigations [9]that "scaling" is not accurate enough for structural forced response analyses because of the phase difference in localunsteady pressures acting on the blade surface. Therefore, the turbine was to be modeled with the true blade ratiosmeaning, in this particular case of prime numbers of blades, a full 3600 CFD model approximately 7 times larger ingrid size.


092407

o

·.

3.5 ,..---------------------------,It:

~~ 3 +-----------------------Ul..~ 2.5 +------------wc

~ 2o~ 1.5

tJ) 1tJ)w~ 0.5tJ)

9 10 11

MODE

12 13

Figure 5. J-2X Mark72-F Fuel Turbine 151 blade alternating stress due to 151 vane excitation.Shown is the ratio between the stress calculated based on a scaled grid CFD solution and the

stress calculated based on full annulus un-scaled CFD solution

Figure 5 documents the forced response analysis results for the 151 blade. The 3600 model shows higheralternating stresses for mode 9, 10, 11 and 12 but lower for mode 13.

A section of the computational grid used in the fluid simulations is shown in Figure 6. The complete grid systemcontained in excess of 56 million grid points. The simulations were run on between 128 and 176 processors of anAltix cluster located at NASA Ames Research Center.

Figure 6. Computational grid for the J-2X Mark72-F fuel turbine

Figure 7 shows instantaneous static pressure contours at midspan of the turbine. Characteristic of mostsupersonic turbines, the largest pressure drop occurs across the nozzle, with relative constant pressure observed inthe fIrst-stage rotor and second stage of the turbine.


092407

Figure 7. Static pressure (psi) at midspan of the J-2X Mark72-F fuel turbine

The relative frame Mach number at midspan of the turbine is shown in Figure 8. This figure highlights both thesupersonic flow regions, as well as the regions of high loss (e.g., separated flow regions). A complex flow structureis observed between the nozzle and fust-stage rotor. Some of the dominant features include: i) expansion wavesemanating from the trailing edge of the nozzle which reflect off the suction surface of the adjacent nozzle andinteract with the nozzle wake further downstream, ii) a strong bow shock in front of the fust-stage rotor whichimpacts the suction surface of the adjacent rotor (causing the boundary layer to separate) and interacts with the wakeand reflected expansion wave from the nozzle, and iii) flow separation on the second-stage rotor.

Figure 8. Relative frame Mach contours at midspan of the ]-2X Mark72-F fuel turbine

Unsteady pressure envelopes for the J-2X fuel turbine are shown Figure 9 and Figure 10. The majority of thepressure drop in the turbine occurs across the nozzle as shown in Figure 9. The flow is choked at the throat of the


092407

·.

, '

nozzle, so the unsteadiness is confined to the uncovered portion of the suction surface and is caused by interactionswith the rotor and the expansion wave from the adjacent nozzle. The greatest unsteadiness on the rotor occurs nearthe leading edge and is generated by several sources, including: a) passing wakes from the upstream nozzles, b)boundary layer separation induced by the bow shock of the adjacent rotor, and c) potential interactions with theupstream nozzles. Overall there is a slight pressure rise across the rotor near the hub and up through midspan.

1.2.--------.----,----r---.---.------.----,----r---.------, 0.7.-----.----,---,---.---r----,r----,-----r---,------,

- Minimum Pressure- Maximum Pressun:- lime-Averaged Pressure

0.5

0.6

- Minimum Pressure- Maximum PrciSurc- Time-Averaged Pressure

0.8

0.4

0.2

~

_Ie: 0.6

~

°o!----'-------,0,l:;.2,.---'---f0.4:----'---,l0.6:------'---.,,0'::-.8--'-~

XfCo0::-----'--,0"'.2,.---'---:-0.4:------'----::0.76--'-----:0~.8---'--------c

XfC

Figure 9. Midspan pressure envelope for the first stage of the J-2X fuel turbine. - nozzle on left,

The unsteady pressure envelopes on the second stage vane and rotor are shown in Figure 10. On the second stagevane the unsteadiness is generated by potential and wake interactions with the upstream and downstream rotors. Ashock wave is evident at approximately 80% axial chord of the vane. A relatively large amount of unsteadiness isevident near the leading edge of the second stage rotor, and is caused by the wakes from the upstream vanes.Overall, there is little change in pressure across the second stage of the turbine.

0.5,...--r----,----,-----,---r--,----,---.----r-----, 0.3,...--,--,----,,.------.----,-----,----,-----,---r--,

oOL-----'--0""'.2:------'----:O.L...--'----'0.-,6--'--0='=.8:---'----'

XfC

0.4

0.1

- Minimum Pre!l!>ure- Maximum Pressure- Time~A\!eraged Pressure

1c:: 0.15

Ci:

o0~----'---!O.2:------'---0-f.4:------'---,0:':.6--'----:O':-.8--'---.JXfC

Figure 10. Midspan pressure envelope for the second stage of the J-2X fuel turbine. - vane onleft, rotor on right. rotor on right.

Forced response analyses indicate high level of alternating stress for the 1st vane 19th mode in the baselineconfiguration. This mode is mainly a complex trailing edge mode excited by the interaction between the expansionwaves at the vane trailing edge and the bow shocks at the blade leading edge. The 19th mode could be eliminatedfrom the operating range by bowing the vane - as shown in Figure 11 - and thus increasing its frequency by 15%.

7American Institute of Aeronautics and Astronautics.

092407

'. I

Nozzle Trailing EdgeSuction Side

Rotation Pressure Side

Figure 11. Bowed nozzles in the J-2X fuel turbine

The same grid topology as the one shown in Figure 6 was used for the simulations including bowed nozzles.The turbine performance is similar for both the baseline and bowed nozzles. Figure 12 and 14 show decompositionsof the unsteady pressure on the nozzle and rotor, respectively, for the two simulations. The unsteadiness on thenozzle actually increases with the bowed nozzle, while the unsteadiness on the rotor decreases slightly.

C 3 C 3

6 6Q)

Q)~ 2.5

~ 2.50u:.. 0u:..>,

2 >,"0 2o:l "0~

o:lCIJ ~:5 1.5

CIJ:5 1.5Q)

"0 Q)

.g 1 "0

.; 10-E 0--< 0.5 E<.,!. -< 05

<; <.,!.

::c <;80000 le+05 ::c

1.2..05

Figure 12. Nozzle unsteadiness with the baseline nozzle (left) and bowed (right) nozzle; rotor-1blade passing frequency = 35,153 Hz.


092407

·. .. ~

G:' 10 G:' 10

;g ;g<U <U

~u

8 ... 8

& 0~

;>, ;>,"0 "0

'" 6 '" 6~ ~

'" '"C C;:J ;:J<U 4 <U 4

"0 "0

a .~i5.. i5..e 2 e 2

<: <:~ ~

-a -a::r: ::r:

80000 1.~1.~ 1.20+05

Figure 13. Rotor-1 unsteadiness with the baseline Oeft) and bowed (right) nozzle; nozzle bladepassing frequency = 18,850 Hz.

Subsequent structural analyses confirm the low level of stress for the 19th mode, solving the problem for thisspecific mode. However, this technique of curving the vanes is not applied to the fmal design because the bowedvane alternating stress for the 7th mode is higher than that of the baseline design. Instead, additional mechanicaldamping is added to lower the vane trailing edge vibrations.

IV. The J-2X Oxidizer Turbine Mark72-0

A. General configuration

The baseline configuration for the J-2X Mark72-0 turbine is the heritage J-2 oxidizer turbine design, also knownunder the Mark15-0 denomination. The J-2 oxidizer turbine is a two stage impulse turbine which operates at lowerpressure ratios than the fuel turbine, and for this reason, whereas the fluid velocity at the nozzle discharge is in thelow supersonic range, the rotor inlet velocity in the relative frame of motion is subsonic.

Optimizations made on the J-2X engine cycle imposed an operating pressure ratio across the Mk72-0 turbinelower than the J-2 engine cycle and consequently the flow regime at the Mark72-0 nozzle discharge is subsonic. Asummary of the turbine operating parameters for the same regime of operation is documented in Table 2.

The new flow path design of the J-2X Mark 72-0 turbine is shown in Figure 14. The turbine elements areidentical to the heritage design with the exception of the 151 nozzle vanes. Similar to the case of the Mark72-Fturbine, the Mark72-0 Campbell diagram indicates possible resonant interference at the 151 stage rotor blades for theheritage configuration and as a design solution the number of nozzle vanes has been changed from 71 to 76.


092407

/

76 vanes 107 blades

I

113 vanes 97 blades

I I

·.

Figure 14. J-2X Mark72-0 oxidizer turbine flow path.

Another specific concern for the Mark72-0 turbine is related to the amount of axial thrust developed by theturbine. With a mean diameter of 15.5 inches, the turbine disks have significant area over which fluid pressure canexercise net thrust. A detailed study of the turbine configuration identified the pressure in the nozzle cavity area ascritical in determining the value of the axial thrust. In tum, at given turbine operating boundary conditions, the maindriver determining the pressure in the nozzle cavity area is the nozzle throat area. Consequently, following designnotes found on 1960's documentation, detailed nozzle throat area measurements were made on heritage hardware.The measurements confirmed the notes which indicate that original 1-2 Mark15-0 production turbines had a throatarea larger by approximately 1.0 in2 than the blue print specification. The airfoils for the new row of 76 nozzles havebeen therefore designed with the same throat area as the baseline production hardware with proven flightperformance. In addition to these precautions, detailed CFD analysis has been made for the flow in the turbinecavity areas, specifically addressing the axial thrust calculations.

Pressure ratioRPMu/CoMean Diameter (in)

2.00110000.17315.5

Table 2: 294K Ibf thrust operating point of the J-2X Mark72-0 oxidizer turbine.

B. J-2X Mark72-0 turbine analyses

The 3D Phantom code coupled with ANSYS finite element forced response analyses are employed to insure thereliability of turbine components similar to the Mark72-F analysis. The turbine has been modeled with the true bladeratios meaning, in this particular case of prime numbers of blades, a full 3600 CFD model comprised of around 58mil1ion grid points. Figure 15 shows a section of the computational grid. The simulations were run on Altix clustersof 128 to 174 processors at the NASA Ames Advanced Supercomputing division (NAS) in Moffet Field, California.


092407

. _.-- -------------

- -------------.---------

-- --

-~---------

-----------

..... - - -- - - .. -- ---------------

'~~~~

::'. ~ ~ ._._~~~==~'iIIf,.~. ,~ --------..:.


092407

Figure 15. Computational grid for the J-2X Mark72-0 oxidizer turbine

Figure 16. Static pressure (psi) and Mach values at midspan of the J-2X Mark72-0 oxidizer

Figure 16 shows instantaneous static pressure and Mach contours at midspan of the turbine. Whereas theflow regime at the nozzle discharge has local transonic regions, the overall flow field remains subsonic. Weakvortex shedding can be also observed at the nozzle trailing edge.

Unsteady pressure envelopes for the J-2X Mark72-0 oxidizer turbine are shown in Figure 17 and Figure 18. Theplots show benign operation without large extremes, smooth pressure contours and good loading on all rows.

..

..

OJ!

U.4

U.2

- Mla,mum ~urt

- Ma..'timum Prtuur'C- Time-A..,:; Prcuure

U.2- Mlaimum Pressure- M::ucimu., Pre.u.ure- TlO*A\>p Prtuure

UO:---'~-:f-O.l::---.L--:O~.4:--""""-~U':-b--'---::o.::-K--'--'.

X/C°0!--'-----:'0.""2--'---,!-OA,---'---'Ob--.L--0"'.Hr--'---'

XlC

Figure 17. Midspan pressure envelope for the first stage of the J-2X Mark72-0 oxidizer turbine.- nozzle on left, rotor on right.

O.ll ,----,,----,,-----,---,---,---,--,--,---,----,

0.1

0.4

0.7,...--r--r----,.---.......,....---,..-----,..-----,..-----,---,---,

0.3 03

1- Mi.inlWU Pn:8o.urc~1- Ma.xinUl1 Pn:nun:::- lime-A\'g Pressure

0.20=--'--~O.':-l--'---::O ..,..~--'---::0.6:----'--0:!-..H::---'--~

XlCO.2n:---''----7-:----'---7'J.4r---'--:f-Ob:----L---:fo.s::---.L-~

XlC

Figure 18. Midspan pressure envelope for the second stage of the J-2X Mark72-0 oxidizerturbine. - stator vane on left, rotor on right.

Given the compact design of the turbine, special attention has been give to the operating axial gaps. Heritagedocumentation indicates repeated design changes on the history of Mark15-0 turbine addressing variations in axialgap during operation. Stacking analysis and correlation with recently finalized power-pack testing have builtconfidence in the configuration chosen for the Mark72-0 design. In addition, CFD studies have been made forassessing the effect of possible gap changes in operation and establish ways to mitigate those effects which are notacceptable. Figure 19 shows such an analysis in which the baseline gaps are modified by a change in the 2nd stagestator vane axial location. The effect of the change is observable in the FFT plots of the unsteady force spectrum forthe 2nd stage rotor: the change in the gap value appears to amplify a low frequency band vibration which mayinterfere with one or more natural frequencies associated with lower modes of the rotor. As a result, a specialassembly procedure has been conceptualized for the pump which ensures operating axial gaps at desired values.


092407

~ ,..

C'"gIRotor-1 I

\ 12xRotor-1 I

I Baseline gaps

IOIIlIl ,.....

IRotor-1 I~/; 12xRotor-11

Q.G.2

'r jr.etalive gep ohoog

E«~

Ci:c

IOIIlIl ,-Figure 19. The effect of axial gap changes between 1st stage rotor and 2nd stage stator and 2nd

stage stator and rotor on the unsteady force spectrum for the 2nd stage rotor.

v. Conclusions

The J-2X turbines which have been denoted as Mark72-F and Mark72-0 for the fuel and oxidizer side,respectively, have been designed by preserving the flight proven heritage designs from the J-2 and J-2S engines.Pratt and Whitney Rocketdyne's (PWR) turbopump design team has applied robust and validated design processeswhile operating jointly with the NASA MSFC team in an efficient and synergistic manner. 3-D unsteady CFDresults obtained by the MSFC team from Phantom CFD simulations run at the NAS center in Moffet Field,California have been analyzed for turbine performance verification by the PWR aerodynamic analysis team. Thesteady and unsteady pressure data from the CFD solution have been applied on the ANSYS structural models forstress and High Cycle Fatigue (HCF) analysis at PWR. The careful analysis has led to minimal design changes ofthe heritage design, specifically the turbine nozzles for both turbines have been redesigned. Various trades haveaddressed concerns related to airfoil HCF, axial thrust developed on the turbine and axial gap effects. The resultingdesigns for the Mark72-F and Mark72-0 turbines are robust, safe, and performing per requirements set by the J-2Xengine cycle.

References

1. W. Sack and 1. Watanabe, "Development Progress of MB-XX Cryogenic Upper Stage Rocket Engine",39th AlAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Huntsville, Alabama, July 20-23,2003, AIAA-2003-4486

2. Marcu, 8., "Turbo-Pump Development and Test for the MB-XX Advanced Upper Stage EngineDemonstrator", 42nd AlAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Sacramento,California, July 9-12, 2006, AIAA-2006-4380.


092407

3. Sondak, D. L. and Dorney, D. J., "General Equation Set Solver for Compressible and IncompressibleTurbomachinery Flows," AIAA 2003-4420, 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference andExhibit, Huntsville, AL, July 20-23, 2003.

4. Baldwin, B. S., and Lomax, H., "Thin Layer Approximation and Algebraic Model for SeparatedTurbulent Flow," AIAA Paper 78-257, Huntsville, AL, January, 1978.

5. Oefelein, 1. C., Sandia Corporation, Livermore, CA, Private Communication, December, 2002.

6. http://webbook.nist.gov/chemistry/fluid

7. Dorney, D. 1., and Sondak, D. L., "Application of the PHANTOM Analysis to Impeller and DiffuserGeometries," 52nd JANNAF Propulsion Meeting, Las Vegas, NV, May 10-13,2004.

8. Starken, H., Yongxing, Z and Schreiber, H-A, "Mass flow limitation of supersonic blade rows due toleading edge blockage" ASME 84-GT-233

9. Clark, 1. P., Stetson, 1. M., Magge, S. S., Ni, R. H., Haldemann, C. W. and Dunn, M. G.: "The Effect ofAirfoil Scaling on the Predicted Unsteady Loading on the Blade of a 1 and Yz Stage Transonic Turbine and aComparison with Experimental Results", ASME Paper No. 2000-GT-0446


092407

...

turbine design and analysis for the j-2xengine turbopumps

Documents